Data transfer
Data transfer is the movement of digital information between memory locations, computing components, systems, or network endpoints.
Quantization and Fast Inference
Data movement is the transfer of data among storage, memory, and computational locations during computation.
- How it is used
- Frames data movement as a central efficiency constraint motivating quantization for inference.
- How deep it goes
- Quantifies energy across memory levels and arithmetic, without detailing transfer mechanisms or runtime implementations.
- Examples
- Compares approximate energy costs for L1, L2, HBM, and INT8 multiply-add operations.
- What you can do after
- Understand why memory movement can constrain inference more than arithmetic.
Read how Quantization and Fast Inference explains data transfer →
CUDA for LLMs
Data movement is the movement of data between locations or memory spaces in a computing system.
- How it is used
- Coordinates CUDA work across separate host and device memory spaces, making explicit transfer part of kernel execution.
- How deep it goes
- Traces allocation, copies, and kernel execution, then links repeated HBM traffic to attention bottlenecks.
- Examples
- Uses vector addition and attention matrices to illustrate transfers between host and device and between memory tiers.
- What you can do after
- Apply CUDA's cudaMalloc, cudaMemcpy, and cudaFree workflow to move vector data through a GPU computation.
Fundamentals of Cybersecurity
Data in transit is data being transmitted between systems, devices, or locations over a communication channel.
- How it is used
- Distinguishes data being transmitted from data at rest and data in use to guide protection choices.
- How deep it goes
- Defines the state, contrasts its exposure with other data states, and connects transmission risks to controls without implementation steps.
- Examples
- Uses file transfers, email, instant messages, bank logins, unsecured Wi-Fi eavesdropping, and on-path message alteration as examples.
- What you can do after
- Evaluate transmission risks and identify encryption or DLP as relevant protections.
Read how Fundamentals of Cybersecurity explains data transfer →
Also covered in
- AI-Powered Developer: Build software with ChatGPT and Copilot
- MongoDB 8.0 in Action, Third Edition
- Privacy-Preserving Machine Learning
- Infrastructure as Code, Patterns and Practices: With examples in Python and Terraform
- Apache Kafka in Action
- Microsoft Azure in Action
- Data-Driven Cybersecurity
- Think Like a CTO
- Data Pipelines with Apache Airflow, Second Edition
What to know first
- encryptionEncryption is the cryptographic transformation of readable data into ciphertext that requires appropriate decryption information to recover the original data.
- availability zoneAn availability zone is an isolated infrastructure location within a cloud region that serves as a distinct failure domain for deployed resources.
- float16float16 is a 16-bit floating-point format, commonly IEEE 754 binary16, with less precision and range than 32-bit floating-point formats.
- fused multiply-addA fused multiply-add is an arithmetic operation that multiplies two values, adds a third, and rounds the result once.
Related concepts
- cudaMemcpyuses · cudaMemcpy is a CUDA Runtime API function that copies data between memory regions, including host and device memory.
- encryptionuses · Encryption is the cryptographic transformation of readable data into ciphertext that requires appropriate decryption information to recover the original data.
- TLSused by · Transport Layer Security (TLS) is a cryptographic protocol that protects network communications through confidentiality, integrity, and authentication of communicating endpoints.
- address spaceuses · An address space is a set or range of addresses available within a computing or communication system for identifying memory locations or other resources.
- data at restcontrasts with · Data at rest is digital information stored on persistent or nonvolatile media rather than actively transmitted across a network.
Questions about data transfer
What is data transfer?
Data transfer is the movement of digital information between memory locations, computing components, systems, or network endpoints.
Which Manning books explain data transfer?
Quantization and Fast Inference, CUDA for LLMs and Fundamentals of Cybersecurity.
What should I know before learning data transfer?
Manning books that teach data transfer usually introduce encryption, availability zone, float16 and fused multiply-add first.
What is the difference between data transfer and data at rest?
Data transfer is the movement of digital information between memory locations, computing components, systems, or network endpoints. Data at rest is digital information stored on persistent or nonvolatile media rather than actively transmitted across a network.
What kinds of data transfer are there?
Kinds of data transfer include CPU-GPU data transfer.
What is data transfer also called?
Data transfer is also called data movement.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.


